Efficient bulk-loading on dynamic metric access methods

This paper presents a new technique and two algorithms to bulk-load data into multi-way dynamic metric access methods, based on the covering radius of representative elements employed to organize data in hierarchical data structures. The proposed algorithms are sample-based, and they always build a valid and height-balanced tree. We compare the proposed algorithm with existing ones, showing the behavior to bulk-load data into the Slim-tree metric access method. After having identified the worst case of our first algorithm, we describe adequate counteractions in an elegant way creating the second algorithm. Experiments performed to evaluate their performance show that our bulk-loading methods build trees faster than the sequential insertion method regarding construction time, and that it also significantly improves search performance. (C) 2009 Elsevier B.V. All rights reserved.

Image indexing and retrieval using an ART-2A neural network architecture

Traditional content-based image retrieval (CBIR) systems use low-level features such as colors, shapes, and textures of images. Although, users make queries based on semantics, which are not easily related to such low-level characteristics. Recent works on CBIR confirm that researchers have been trying to map visual low-level characteristics and high-level semantics. The relation between low-level characteristics and image textual information has motivated this article which proposes a model for automatic classification and categorization of words associated to images. This proposal considers a self-organizing neural network architecture, which classifies textual information without previous learning. Experimental results compare the performance results of the text-based approach to an image retrieval system based on low-level features. (c) 2008 Wiley Periodicals, Inc.

Adhesion to Er : YAG laser-prepared dentin after long-term water storage and thermocycling

This in vitro study evaluated the microtensile bond strength of a resin composite to Er:YAG-prepared dentin after long-term storage and thermocycling. Eighty bovine incisors were selected and their roots removed. The crowns were ground to expose superficial dentin. The samples were randomly divided according to cavity preparation method (I-Er:YAG laser and II-carbide bur). Subsequently, an etch & rinse adhesive system was applied and the samples were restored with a resin composite. The samples were subdivided according to time of water storage (WS)/number of thermocycles (TC) performed: A) 24 hours WS/no TC; B) 7 days WS/500 TC; C) 1 month WS/2,000 TC; D) 6 months WS/12,000 TC. The teeth were sectioned in sticks with a cross-sectional area of 1.0-mm(2), which were loaded in tension in a universal testing machine. The data were subjected to two-way ANOVA, Scheffe and Fisher`s tests at a 5% level. In general, the bur-prepared group displayed higher microtensile bond strength values than the laser-treated group. Based on one-month water storage and 2,000 thermocycles, the performance of the tested adhesive system to Er:YAG-laser irradiated dentin was negatively affected (Group IC), while adhesion of the bur-prepared group decreased only within six months of water storage combined with 12,000 thermocycles (Group IID). It may be concluded that adhesion to the Er:YAG laser cavity preparation was more affected by the methods used for simulating degradation of the adhesive interface.